Towards Multi-Agentic AI for automated software design and modelling: challenges and opportunities

Agentic AI has recently emerged as a new paradigm where autonomous AI agents, empowered with Large Language Model capabilities, are capable of perceiving, reasoning and acting independently to pursue goals. In this paper, we explore the key challenges in developing Agentic AI for automated software design and modelling. These challenges include translating natural language requirements into design models, resolving ambiguities, maintaining consistency within and between design models, and managing conflicts and inconsistencies when merging different versions of a design model. To address those challenges, we propose a conceptual Multi-Agentic AI framework in which autonomous, goal-driven AI agents capable of interpreting requirements, generating design artefacts and collaborating on software design models.